Estimation of heat transfer performance on mixed convection in an enclosure with an inner cylinder using an artificial neural network

Hyun Woo Cho, Young Min Seo, Yong Gap Park, Sudhanshu Pandey*, Man Yeong Ha* (Corresponding Author)

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

7 Citations (Scopus)

Abstract

The effect of variations in the aspect ratio (AR) of an inner cylinder on 3D mixed convection is examined by numerical means. Three Reynolds numbers are considered, Re = 100, 500, and 1000, with a fixed Grashof number of 105 and a Prandtl number of 0.7. The flow patterns formed inside the enclosure are categorized as spanwise, streamwise, and clockwise for the given Reynolds numbers. The rate of heat transfer was enhanced by 16.8% and 20.6% at Re = 100 and 500 for AR = 4, respectively, compared with the rate of heat of transfer at AR = 1 and AR = 0.5, respectively. Similarly, the rate of heat transfer was increased by 14.0% at Re = 1000 for AR = 1 compared with the rate of heat transfer at AR = 0.25. These results demonstrate that the heat transfer characteristics of mixed convection can be predicted readily using an ANN based on the data set created from a few cases of direct numerical simulation (DNS); the same results would otherwise require days to calculate using DNS. Furthermore, the values predicted by the ANN are in good agreement with the data obtained by DNS. The heat transfer is affected by Reynolds number, aspect ratio, radius of the cylinder, and dimensionality of the simulation, in that order.

Original languageEnglish
Article number101595
JournalCase Studies in Thermal Engineering
Volume28
DOIs
Publication statusPublished - Dec 2021
MoE publication typeA1 Journal article-refereed

Keywords

  • Artificial neural network
  • Aspect ratio
  • Inner cylinder
  • Three-dimensional mixed convection

Fingerprint

Dive into the research topics of 'Estimation of heat transfer performance on mixed convection in an enclosure with an inner cylinder using an artificial neural network'. Together they form a unique fingerprint.

Cite this